zynQ

AI content creator | AI news, agents & tools | daily visual breakdowns

[Milo @m1lo_exe]

STOP REWRITING YOUR F*CKING PROMPTS

Your model isn't the problem. the room you put it in is

Same weights, two rooms:
> empty chat box → casual prose
> repo with tests, a project map and a terminal → working software

That room is the harness, and it runs in 3 layers:
→ harness layer - the model is the CPU, the context window is the RAM. it reads one file from a map, never a 20-page doc dump
→ loop layer - no "are you sure?". tests, linters and schema checks hand back raw output. 4 attempts max, then a human
→ graph layer - a planner, a builder on an isolated branch and an adversarial reviewer. nobody carries the whole project in one thread

The rule that makes it hold is dual-encoding:
1/ say "never delete tests" in the prompt
2/ make rm -rf throw EPERM at the OS level

The prompt guides. the gate protects

The full tree :
> HARNESS.md - deliverable, constraints, non-goals, acceptance
> project map - /src, /tests, /docs/decisions.md
> CLAUDE.md - memory that survives resets and model swaps
> sensors - lint, tests, schema before anything counts as done
> escalation - same error twice = stop and ask

PASTE THIS INTO YOUR CLAUDE CODE ↓

Read my repo and write a HARNESS.md for it:
1. Bounded contract: deliverable, constraints, non-goals, acceptance criteria
2. Project map: where docs, schemas and tests live
3. Verification protocol: lint → tests → schema check, in that order
4. Escalation: if a test fails twice on the same error, stop and show me the log
5. Memory: move durable decisions into CLAUDE.md
6. Show me the file as a diff first. No edits until I say go

New model next month?
> your prompts break
> your harness doesn't

Save it, it'll come in handy

[zynQ @zynQ6969]

𝗧𝗛𝗜𝗦 𝗜𝗦 𝗙*𝗖𝗞𝗜𝗡𝗚 𝗜𝗡𝗦𝗔𝗡𝗘

two AI labs just broke a 27-year-old rule of computer science. on the same day

Claude did it with zero human input

𝘄𝗵𝗮𝘁 𝗵𝗮𝗽𝗽𝗲𝗻𝗲𝗱
→ oct 16: an internal Claude model found the core algorithm. 14M output tokens, no human input
→ same day: an OpenAI model found its own disproof. Lean-checked oct 17
→ oct 26: both announced, hours apart. working independently

𝘄𝗵𝗮𝘁 𝘁𝗵𝗲𝘆 𝗯𝗿𝗼𝗸𝗲
the Exponential Time Hypothesis (1999). it says 3-SAT, the classic "hard puzzle" of CS, can't be solved meaningfully faster than brute force

Anthropic's paper: "A 2^o(n) time algorithm for 3-SAT"
author: "Claude, Anthropic"

𝗵𝗼𝘄 𝗶𝘁'𝘀 𝗰𝗵𝗲𝗰𝗸𝗲𝗱
• 1,050 Lean modules, verified by the Lean kernel + an independent checker
• the appendix says no human has checked the full proof yet
• OpenAI's version goes to a UCSD + MIT group, led by the two people who proposed ETH in 1999

imagine checking the AI proof that kills your own hypothesis

𝘁𝗵𝗲 𝗳𝗮𝗹𝗹𝗼𝘂𝘁
Claude's own literature search found 800+ papers resting on ETH-style assumptions. those results now need new foundations

𝘁𝗵𝗲 𝗰𝗮𝘁𝗰𝗵 𝗻𝗼𝗯𝗼𝗱𝘆'𝘀 𝗽𝗼𝘀𝘁𝗶𝗻𝗴
the new algorithm only beats brute force past ~10^10,000 variables. there are ~10^80 atoms in the universe
→ no practical speedup
→ not P = NP
→ nothing changes for your laptop. yet

historic for math. useless for your laptop. for now

the real headline: AI now writes proofs faster than humans can check them

would you trust a proof no human has read? ↓

[zynQ @zynQ6969]

𝗧𝗛𝗜𝗦 𝗜𝗦 𝗙*𝗖𝗞𝗜𝗡𝗚 𝗜𝗡𝗦𝗔𝗡𝗘

two AI labs just broke a 27-year-old rule of computer science. on the same day

Claude did it with zero human input

𝘄𝗵𝗮𝘁 𝗵𝗮𝗽𝗽𝗲𝗻𝗲𝗱
→ oct 16: an internal Claude model found the core algorithm. 14M output tokens, no human input
→ same day: an OpenAI model found its own disproof. Lean-checked oct 17
→ oct 26: both announced, hours apart. working independently

𝘄𝗵𝗮𝘁 𝘁𝗵𝗲𝘆 𝗯𝗿𝗼𝗸𝗲
the Exponential Time Hypothesis (1999). it says 3-SAT, the classic "hard puzzle" of CS, can't be solved meaningfully faster than brute force

Anthropic's paper: "A 2^o(n) time algorithm for 3-SAT"
author: "Claude, Anthropic"

𝗵𝗼𝘄 𝗶𝘁'𝘀 𝗰𝗵𝗲𝗰𝗸𝗲𝗱
• 1,050 Lean modules, verified by the Lean kernel + an independent checker
• the appendix says no human has checked the full proof yet
• OpenAI's version goes to a UCSD + MIT group, led by the two people who proposed ETH in 1999

imagine checking the AI proof that kills your own hypothesis

𝘁𝗵𝗲 𝗳𝗮𝗹𝗹𝗼𝘂𝘁
Claude's own literature search found 800+ papers resting on ETH-style assumptions. those results now need new foundations

𝘁𝗵𝗲 𝗰𝗮𝘁𝗰𝗵 𝗻𝗼𝗯𝗼𝗱𝘆'𝘀 𝗽𝗼𝘀𝘁𝗶𝗻𝗴
the new algorithm only beats brute force past ~10^10,000 variables. there are ~10^80 atoms in the universe
→ no practical speedup
→ not P = NP
→ nothing changes for your laptop. yet

historic for math. useless for your laptop. for now

the real headline: AI now writes proofs faster than humans can check them

would you trust a proof no human has read? ↓

[levent @__alpoge__]

“The core algorithm was found on October 16 by an internal Claude research model with 14M output tokens and no human input. Anthropic researchers worked with Claude to improve the presentation, resulting in this paper.”

https://t.co/rZLajqbVH8

[Milo @m1lo_exe]

HOLY F*CK, your Claude loop is spending more on deciding than on building

600 tiny yes-or-no calls a night. $765 a month just to ask "is this safe" and "are we done"

One 100 ms model takes the same loop to $3

Leave a loop running overnight and read the transcript in the morning. most turns aren't writing code. they're asking permission:

> is this command safe
> which file is next
> did that test output mean anything
> is the goal met or should it keep going

Every one of those goes to the same model that writes your code. about 4 cents a question on Fable 5.1, and it takes seconds

The fix is 3 hooks:
> safety gate - a PreToolUse hook gives every bash command a risk label before it runs. read-only passes, destructive gets blocked, anything external comes back to you
> stop hook - Claude can't call itself done. https://t.co/6jNXefiEKx prints the evidence, Jev returns the odds the goal is met. above 0.9 it stops, below 0.5 it keeps going
> trigger - a cron job triages new issues, and Claude only wakes up for real bugs

The numbers, 30 nights:
> Fable 5.1 deciding - $765
> Sonnet 5 deciding - $229.50
> Jev deciding - $3.02, about $1.08 if you ask all 3 questions in one call

The part almost nobody builds: hard rules stay in code. force pushes and piped installs get blocked before Jev ever sees them

Claude builds. Jev decides. the cheapest Claude turn is the one that never needed to start

Full code for all 3 hooks is in the article below. bookmark it before your next overnight run

[zynQ @zynQ6969]

@testingcatalog watching this one too. if they show anything on cheaper long-running agent runs that's the real news for builders, the models are already good enough

[zynQ @zynQ6969]

@Ondo tokenized pre-IPO exposure to OpenAI trading 24/7 is wild. the part I'd want to understand: with no public price, what does the note actually track between funding rounds?

[zynQ @zynQ6969]

@injective pricing compute as a market makes sense, GPU hours are basically a commodity now. curious how the index weights chip generations though, an H100 hour and a B200 hour aren't the same product

[zynQ @zynQ6969]

𝗧𝗛𝗜𝗦 𝗜𝗦 𝗔𝗖𝗧𝗨𝗔𝗟𝗟𝗬 𝗙*𝗖𝗞𝗜𝗡𝗚 𝗜𝗡𝗦𝗔𝗡𝗘

Apple's "Pro and tell" event is on oct 28 and everyone expects the new MacBook Pro. people are already sharing AI rebuilds of the leak

so I had Claude rebuild it too. then I did the part nobody did: checked every leak against what Apple actually said

231 parts. 77 keys with legends. lid opens to 112°. 2 of 8 claims confirmed

the trick: every part is code, and every claim gets a tag before it's allowed into the build

𝘄𝗵𝗮𝘁'𝘀 𝗶𝗻𝘀𝗶𝗱𝗲 𝘁𝗵𝗲 𝗿𝗲𝗯𝘂𝗶𝗹𝗱
• space black aluminium, 313 × 221 mm (today's 14" footprint)
• logic board with an "M5 Pro?" chip, vapor chamber, 2 copper heat pipes
• 2 fans, 6 battery cells, speakers, MagSafe + ports
• 77 keys, trackpad, hinge, display cables through the hinge

𝘁𝗵𝗲 𝗹𝗲𝗮𝗸 𝗰𝗵𝗲𝗰𝗸
→ "Pro and tell" event: CONFIRMED (Apple invite)
→ oct 28, Los Angeles: CONFIRMED (Apple invite)
→ OLED display: rumored
→ touchscreen: rumored
→ Dynamic Island: conflicting, some reports say punch-hole
→ M5 Pro / M5 Max: rumored
→ thinner + lighter: rumored, no numbers yet
→ 6 cells, 2 fans: pure guess. mine included

Apple confirmed when and where. not a single spec

0 errors on a rumor is still a rumor

building from leaks takes AI minutes now. the real skill is knowing which leaks are real

paste this into Claude ↓

"Rebuild [any unreleased device] from leaks, in code:
1. List every spec. Tag each one: confirmed by the company, reported, rumored or guess. Cite the source.
2. Build from confirmed + reported specs. Mark the rest as placeholders.
3. Add parts one at a time. Check clearances after each one.
4. End with a table: claim → tag → source.
No code until I say 'go'."

works on any launch: phones, consoles, headsets

I'll rerun this against the real thing on oct 28. save this ↓

[zynQ @zynQ6969]

𝗧𝗛𝗜𝗦 𝗜𝗦 𝗔𝗖𝗧𝗨𝗔𝗟𝗟𝗬 𝗙*𝗖𝗞𝗜𝗡𝗚 𝗜𝗡𝗦𝗔𝗡𝗘

Apple's "Pro and tell" event is on oct 28 and everyone expects the new MacBook Pro. people are already sharing AI rebuilds of the leak

so I had Claude rebuild it too. then I did the part nobody did: checked every leak against what Apple actually said

231 parts. 77 keys with legends. lid opens to 112°. 2 of 8 claims confirmed

the trick: every part is code, and every claim gets a tag before it's allowed into the build

𝘄𝗵𝗮𝘁'𝘀 𝗶𝗻𝘀𝗶𝗱𝗲 𝘁𝗵𝗲 𝗿𝗲𝗯𝘂𝗶𝗹𝗱
• space black aluminium, 313 × 221 mm (today's 14" footprint)
• logic board with an "M5 Pro?" chip, vapor chamber, 2 copper heat pipes
• 2 fans, 6 battery cells, speakers, MagSafe + ports
• 77 keys, trackpad, hinge, display cables through the hinge

𝘁𝗵𝗲 𝗹𝗲𝗮𝗸 𝗰𝗵𝗲𝗰𝗸
→ "Pro and tell" event: CONFIRMED (Apple invite)
→ oct 28, Los Angeles: CONFIRMED (Apple invite)
→ OLED display: rumored
→ touchscreen: rumored
→ Dynamic Island: conflicting, some reports say punch-hole
→ M5 Pro / M5 Max: rumored
→ thinner + lighter: rumored, no numbers yet
→ 6 cells, 2 fans: pure guess. mine included

Apple confirmed when and where. not a single spec

0 errors on a rumor is still a rumor

building from leaks takes AI minutes now. the real skill is knowing which leaks are real

paste this into Claude ↓

"Rebuild [any unreleased device] from leaks, in code:
1. List every spec. Tag each one: confirmed by the company, reported, rumored or guess. Cite the source.
2. Build from confirmed + reported specs. Mark the rest as placeholders.
3. Add parts one at a time. Check clearances after each one.
4. End with a table: claim → tag → source.
No code until I say 'go'."

works on any launch: phones, consoles, headsets

I'll rerun this against the real thing on oct 28. save this ↓

[Superior @andreysuperior]

THIS IS F*CKING INSANE

Apple shows the new MacBook Pro on Oct 28. Opus 5.5 rebuilt the leaked one first: all 126 parts, and the lid opens.

335k triangles, 259 design rules, 0 errors left.

The trick: every part is code, and nothing goes in until its check passes.

what's inside
> tandem OLED touchscreen with the Dynamic Island cut-out
> M5 Pro / M5 Max logic board, unified memory on the chip
> 6 battery cells under the keyboard and palm rest
> 2 fans, a vapor plate and 2 heat pipes over the chip
> six speakers, MagSafe, keyboard, trackpad
> the hinge, with the display cables running through it

the checks that gate every part
→ clearance ≥ 0.05 mm, 0 clashes. tightest spot: the lid sits 0.14 mm over the keys
→ the fans get a clear air path past the heat pipes
→ display cables bend ≥ 0.8 mm in the hinge (these bend at 4 mm). that cable is what killed 2016 MacBook screens (flexgate)
→ 259 design rules pass, then the lid opens live to 112°

it caught a real one: the first heat pipe route blocked the fan outlet by 0.60 mm. it moved the pipe 1.5 mm and the air path cleared by 0.90 mm.

The guide I'm quoting rebuilds software with REA and calls this check the oracle. Same loop, pointed at a laptop Apple hasn't even shown yet.

paste this into Claude Code ↓

"Rebuild [any device] from its leaks, in code:
1. List the specs. Mark each one confirmed or rumored.
2. Write the parts list with dimensions. One function per part.
3. Place parts one by one. After each, check clearance, clashes, cable bends and airflow. Fix every fail before the next part.
4. Animate the moving parts last.
Show me the specs and the parts list first. No code until I say go."

Works on anything with a hinge or a fan: consoles, cameras, drones.

save it for Oct 28.

[Milo @m1lo_exe]

HOLY F*CK, Google explained why most AI agents break in production.

Annie Wang is a software engineer at Google Cloud. she sat down and built an AI agent that tests, scores and fixes itself, live, start to finish. 95 minutes, free

I watched all of it. these are the 5 moments worth your time:

12% → 11:10 - a self-improving agent is 5 boxes. 2 of them are yours
30% → 28:07 - both agents got it right. only the trajectory shows which one is right again tomorrow
55% → 52:21 - 3 ways to score an answer. only one needs no answer key
81% → 1:16:44 - the flywheel + the 3 exits you have to build
88% → 1:23:31 - 2 telemetry pipes. only one feeds the next exam

My favorite part is 28:07. two agents book the same dinner. one checks the calendar first. the other guesses the time and gets lucky. grade only the result and they look identical

The loop isn't the hard part. deciding what "good enough" means is

Learn the harness from Google, then build the full graph below:

[Milo @m1lo_exe]

THIS IS F*CKING INSANE

Someone turned Claude Code into a full quant research team. five agents instead of one chat, each with one job, and one of them exists only to kill your results

Ask a single chat if a trading rule works and it answers in 4 seconds. it never saw the chart. it never ran the test

Here’s what you’re looking at in the video:

→ left column: the whole pipeline. question → scout → backtest → verdict → sweep → skeptic → ledger → paper → runner

→ orange box: Opus 5.5 as the lead. it plans and hands work to the team

→ green box: the actual backtest. 200-day moving average rule on the S&P 500, 1999 to 2026, vs buy and hold

→ blue cards: the three attacks the skeptic runs before you ever see a result

→ bottom: everything gets written to research.md so the team never tests a dead idea twice

the team:
lead (Opus 5.5): turns a question into a plan
scout (Sonnet 5.5): checks the data before anything runs
sweeper (Sonnet 5.5): runs every variation, never just one
skeptic (Opus 5.5): tries to break every finding
librarian (Haiku 5.5): pulls strategies with evidence behind them

the numbers from the published run:
same growth. 6.8% a year vs 6.8%
worst fall -24.4% vs -56.8%
sharpe 0.61 vs 0.44
$10K → $61,526 vs $61,009
42 trades, gross of costs

Same money. less than half the crash

The strongest model goes to the skeptic, not the idea guy. ideas are cheap. killing bad ones is where the money is save and every study ends in one of three fixed verdicts: DATA VALIDATED, TRADE-OFF or NO EDGE. tested on years the rules never saw (after 2015 by default). no agent can invent a sharpe ratio it never calculated

NO EDGE still counts as a result. log it, never retest it without a new reason

this is not a money printer. no costs, no taxes, no slippage in the backtest. paper first, always

full setup with every agent file in the article below ↓

[zynQ @zynQ6969]

𝗵𝗼𝗹𝘆 𝘀𝗵*𝘁. 𝗷𝗮𝗰𝗸 𝗱𝗼𝗿𝘀𝗲𝘆'𝘀 𝗰𝗼𝗺𝗽𝗮𝗻𝘆 𝗷𝘂𝘀𝘁 𝗯𝘂𝗶𝗹𝘁 𝗮 𝘃𝗼𝗶𝗰𝗲 𝗔𝗜 𝘁𝗵𝗮𝘁 𝗻𝗲𝘃𝗲𝗿 𝗿𝗲𝗮𝗱 𝗮 𝘀𝗶𝗻𝗴𝗹𝗲 𝘄𝗼𝗿𝗱

no text. no ChatGPT-style brain underneath. it learned to talk the way you did, by listening to people

it's called Dessa, from Block (yes, the Cash App people)

𝘄𝗵𝘆 𝗶𝘁 𝗳𝗲𝗲𝗹𝘀 𝗱𝗶𝗳𝗳𝗲𝗿𝗲𝗻𝘁:
→ normal voice AI waits for you to finish, thinks, then answers
→ Dessa listens and talks at the same time
→ it says "mm-hm" while you're mid-sentence, cuts in, laughs with you
→ Block says it picks up sarcasm, hesitation and tone

𝘁𝗵𝗲 𝗽𝗮𝗿𝘁 𝘁𝗵𝗮𝘁 𝗯𝗿𝗼𝗸𝗲 𝗺𝘆 𝗯𝗿𝗮𝗶𝗻:
→ reportedly trained on ONE GB200 NVL72 rack. 72 GPUs. the exact rack from my chip teardown
→ reportedly small enough to run on a MacBook Pro with an M4 Max, faster than real time

meanwhile everyone else is burning entire data centers

𝗯𝘂𝘁 𝗯𝗲𝗳𝗼𝗿𝗲 𝘆𝗼𝘂 𝗹𝗼𝘀𝗲 𝗶𝘁:
→ it's a research preview, waitlist only
→ no benchmarks, no parameter count, no weights
→ The Verge said some demo chats felt "a little awkward and uncanny"

so the real question isn't "is it smart"
it's: would you even notice you're talking to a machine? ↓

[zynQ @zynQ6969]

@ouerbianhuaidan 100%. everyone demos the prompt, nobody demos the phone dying at 3am. what killed it most, battery or network drops?

[zynQ @zynQ6969]

𝗵𝗼𝗹𝘆 𝘀𝗵*𝘁. 𝗲𝘃𝗲𝗿𝘆𝗼𝗻𝗲’𝘀 𝗳𝗿𝗲𝗮𝗸𝗶𝗻𝗴 𝗼𝘂𝘁 𝗼𝘃𝗲𝗿 𝗵𝗼𝘄 𝗵𝘂𝗺𝗮𝗻 𝗗𝗲𝘀𝘀𝗮 𝘀𝗼𝘂𝗻𝗱𝘀

it listens while it talks. cuts in, says “mm-hm”, laughs at the right moment
no awkward pause, no robotic turn-taking
but there are 2 problems nobody’s talking about

𝗽𝗿𝗼𝗯𝗹𝗲𝗺 𝟭: 𝗶𝘁 𝗰𝗮𝗻’𝘁 𝗮𝗰𝘁𝘂𝗮𝗹𝗹𝘆 𝗱𝗼 𝗮𝗻𝘆𝘁𝗵𝗶𝗻𝗴 (𝘆𝗲𝘁)
→ from what Block has shown, no tool calls
→ can’t search, book, pay or open your apps
a world-class talker that can’t lift a finger

𝗽𝗿𝗼𝗯𝗹𝗲𝗺 𝟮: 𝘆𝗼𝘂 𝗰𝗮𝗻’𝘁 𝘁𝗲𝗹𝗹 𝗶𝘁’𝘀 𝗔𝗜
→ a voice on the phone used to be proof. not anymore
→ Dessa is still a waitlisted preview, but voice cloning tools are already out there
set this up with your family today:
→ pick a code word
→ hang up and call back on a number you already have
→ urgency + money = red flag
→ verify on another channel
would your parents catch it? ↓

[Kun Chen @kunchenguid]

a lot of people were impressed by dessa because it’s truly giving a next level voice conversation experience

it’s very impressive indeed - you can listen to the demos on the site and feel the magic yourself

but am i the only one seeing the massive red flags here?

1. this model can’t make tool calls.. that instantly kills almost all practical use cases, right?

2. is it net good or net bad to make AI indistinguishable from human?

https://t.co/HNKTZ78aNC

[zynQ @zynQ6969]

𝗵𝗼𝗹𝘆 𝘀𝗵*𝘁. 𝗲𝘃𝗲𝗿𝘆𝗼𝗻𝗲’𝘀 𝗳𝗿𝗲𝗮𝗸𝗶𝗻𝗴 𝗼𝘂𝘁 𝗼𝘃𝗲𝗿 𝗵𝗼𝘄 𝗵𝘂𝗺𝗮𝗻 𝗗𝗲𝘀𝘀𝗮 𝘀𝗼𝘂𝗻𝗱𝘀

it listens while it talks. cuts in, says “mm-hm”, laughs at the right moment
no awkward pause, no robotic turn-taking
but there are 2 problems nobody’s talking about

𝗽𝗿𝗼𝗯𝗹𝗲𝗺 𝟭: 𝗶𝘁 𝗰𝗮𝗻’𝘁 𝗮𝗰𝘁𝘂𝗮𝗹𝗹𝘆 𝗱𝗼 𝗮𝗻𝘆𝘁𝗵𝗶𝗻𝗴 (𝘆𝗲𝘁)
→ from what Block has shown, no tool calls
→ can’t search, book, pay or open your apps
a world-class talker that can’t lift a finger

𝗽𝗿𝗼𝗯𝗹𝗲𝗺 𝟮: 𝘆𝗼𝘂 𝗰𝗮𝗻’𝘁 𝘁𝗲𝗹𝗹 𝗶𝘁’𝘀 𝗔𝗜
→ a voice on the phone used to be proof. not anymore
→ Dessa is still a waitlisted preview, but voice cloning tools are already out there
set this up with your family today:
→ pick a code word
→ hang up and call back on a number you already have
→ urgency + money = red flag
→ verify on another channel
would your parents catch it? ↓

[zynQ @zynQ6969]

@martinvars The part people skip: it was reportedly trained on a single GB200 NVL72 rack. If audio-only training is that efficient, voice AI stops being a big-lab game. Small teams could build their own.

[zynQ @zynQ6969]

@akshay_pachaar Sounding human might be solved. Being useful is the open question. With no text in the loop, how does it look things up, follow instructions or call tools? Curious whether Block pairs it with a text model later or keeps it pure audio.

[zynQ @zynQ6969]

@kimmonismus The local part is the real story. A full-duplex voice model on a laptop means no round trip to a server, so the interruptions and "mm-hm"s can land in real time. Cloud voice AI has to fight network lag before it can even sound human.

[zynQ @zynQ6969]

𝗵𝗼𝗹𝘆 𝘀𝗵*𝘁. 𝗷𝗮𝗰𝗸 𝗱𝗼𝗿𝘀𝗲𝘆'𝘀 𝗰𝗼𝗺𝗽𝗮𝗻𝘆 𝗷𝘂𝘀𝘁 𝗯𝘂𝗶𝗹𝘁 𝗮 𝘃𝗼𝗶𝗰𝗲 𝗔𝗜 𝘁𝗵𝗮𝘁 𝗻𝗲𝘃𝗲𝗿 𝗿𝗲𝗮𝗱 𝗮 𝘀𝗶𝗻𝗴𝗹𝗲 𝘄𝗼𝗿𝗱

no text. no ChatGPT-style brain underneath. it learned to talk the way you did, by listening to people

it's called Dessa, from Block (yes, the Cash App people)

𝘄𝗵𝘆 𝗶𝘁 𝗳𝗲𝗲𝗹𝘀 𝗱𝗶𝗳𝗳𝗲𝗿𝗲𝗻𝘁:
→ normal voice AI waits for you to finish, thinks, then answers
→ Dessa listens and talks at the same time
→ it says "mm-hm" while you're mid-sentence, cuts in, laughs with you
→ Block says it picks up sarcasm, hesitation and tone

𝘁𝗵𝗲 𝗽𝗮𝗿𝘁 𝘁𝗵𝗮𝘁 𝗯𝗿𝗼𝗸𝗲 𝗺𝘆 𝗯𝗿𝗮𝗶𝗻:
→ reportedly trained on ONE GB200 NVL72 rack. 72 GPUs. the exact rack from my chip teardown
→ reportedly small enough to run on a MacBook Pro with an M4 Max, faster than real time

meanwhile everyone else is burning entire data centers

𝗯𝘂𝘁 𝗯𝗲𝗳𝗼𝗿𝗲 𝘆𝗼𝘂 𝗹𝗼𝘀𝗲 𝗶𝘁:
→ it's a research preview, waitlist only
→ no benchmarks, no parameter count, no weights
→ The Verge said some demo chats felt "a little awkward and uncanny"

so the real question isn't "is it smart"
it's: would you even notice you're talking to a machine? ↓

[jack @jack]

We believe voice is the most natural way to interact with intelligence, so we created Dessa, the most authentic full-duplex speech model ever built. Talking with it feels like talking with a person…full of emotion, nuance, pauses, and a natural cadence.

https://t.co/r0qFLdvd6c

[zynQ @zynQ6969]

@Brandel0r Full truth

[zynQ @zynQ6969]

𝗛𝗼𝗹𝘆 𝘀𝗵*𝘁, 𝟰𝟬,𝟬𝟬𝟬 𝗽𝗲𝗼𝗽𝗹𝗲 𝗷𝘂𝘀𝘁 𝘄𝗮𝘁𝗰𝗵𝗲𝗱 𝗮 "𝗖𝗹𝗮𝘂𝗱𝗲 𝗖𝗼𝗱𝗲 𝗯𝗼𝘁 𝘁𝘂𝗿𝗻𝗲𝗱 $𝟲𝟴 𝗶𝗻𝘁𝗼 $𝟳𝟱𝟬𝗞" 𝗽𝗼𝘀𝘁 𝗮𝗻𝗱 𝗻𝗼𝗯𝗼𝗱𝘆 𝗱𝗶𝗱 𝘁𝗵𝗲 𝗺𝗮𝘁𝗵.

So I did.

𝗧𝗵𝗲 𝗰𝗹𝗮𝗶𝗺:
→ $68 to $750,000 is 11,029x
→ That's doubling your money every single day for 13+ days
→ With zero losing days, on a bot built in 48 hours

𝗣𝗿𝗼𝗯𝗹𝗲𝗺 𝟭, 𝘁𝗵𝗲 "𝗮𝗿𝗯𝗶𝘁𝗿𝗮𝗴𝗲":
→ Price gaps between big crypto markets are usually a few hundredths of a percent
→ Standard exchange fees are around 0.1% to buy and 0.1% to sell
→ So most gaps are smaller than the fee you pay to take them

𝗣𝗿𝗼𝗯𝗹𝗲𝗺 𝟮, 𝘀𝗽𝗲𝗲𝗱:
→ Pro trading firms put their servers right next to the exchange
→ They react in microseconds
→ A bot on a laptop or iPad over home Wi-Fi shows up 100+ ms later
→ By then the gap is almost always gone

𝗣𝗿𝗼𝗯𝗹𝗲𝗺 𝟯, 𝘁𝗵𝗲 𝗮𝘀𝗸:
"Comment Claude, like, repost, follow, and I'll DM you the setup."
Every comment pushes the post to more people. The engagement is the product, not the bot.

𝗪𝗵𝗮𝘁 𝗖𝗹𝗮𝘂𝗱𝗲 𝗖𝗼𝗱𝗲 𝗶𝘀 𝗮𝗰𝘁𝘂𝗮𝗹𝗹𝘆 𝗴𝗿𝗲𝗮𝘁 𝗳𝗼𝗿 𝗵𝗲𝗿𝗲:
→ Backtesting a strategy on real data before you risk a cent
→ Building dashboards and alerts
→ Finding out your edge disappears after fees, for free

If it printed $750K, why would anyone sell it for a comment?

[Luis Enzo @techluisenzo]

A 19-year-old Japanese student built a trading bot with Claude Code in 2 days.

Used his iPad as a second monitor.

First night: $6,732 profit.
Starting capital: $68.

Total profit so far: $750,000.

Here's how it works:

The bot scans over 50 markets simultaneously.

Syncs live BTC data from Binance every second.

Spots price errors before humans even notice.

The edge is pure speed + pattern recognition.

While traders stare at charts trying to predict the next move, his bot is already executing on mispricing across dozens of markets.

No guessing.
No emotions.
No hesitation.

Just Claude Code logic finding gaps that close in seconds.

He built the entire system in 48 hours:

→ Claude Code handles the trading logic
→ Binance API feeds real-time BTC data
→ iPad displays multi-market monitoring
→ Executes trades when arbitrage windows open

The system runs 24/7.

Every price dislocation = profit opportunity.

Most people are still trading manually, refreshing charts, second-guessing entries.

Meanwhile this 19-year-old engineering student turned $68 into $750K by letting Claude Code do what humans can't: process 50 markets instantly and execute without fear.

Why are people still trading manually?

💡I'm giving away the exact Claude Code setup for free.

24 hours only.

To get it: 👇

1️⃣ Comment "Claude"
2️⃣ Like Comment and Repost

Follow @techluisenzo

I'll DM you the complete setup.